ARC122

Overview

In this lab you will learn the fundamentals of prompt engineering using Generative AI and Google Cloud. During the lab you will have access to a chatbot that will act as a .

If you are new to Generative AI or looking for an overview of how to get started, you are in the right place. Read on to learn about the specifics of this lab and areas that you will get hands-on practice with.

In this lab learn use a chat application to interact and learn:

• The basics of prompt engineering
• Why context in relation to AI is important
• How to work with Generative AI

Prerequisites

Over the course of this lab the following elements are required:

• Vertex AI
• Generative AI models

Task 1. Access the Chat Application

Open the to gain access to the lab chat application.

Note: The chat bot link works in both a normal browser tab and an incognito window. An initial loading screen will appear while the lab data is being prepared.

From here you will be able to interact with a Generative AI model during the course of this lab.

Note: The above image is the main chat screen. The chat screen can include multiple personas reflecting different knowledge domains. Each persona has specific but limited abilities to demonstrate the functionality of generative ai on a defined topic.

The lab mimics a chat application interface to provide access to a generative ai model. Select a persona from the contacts list to initiate a discussion on a specified topic.

Note: The application uses a simplified version of the chat/bison-001 model. In this demo, each query is distinct and therefore, the user must input the full context per query.

The application allows the user to perform interactions with the generative ai from Google Cloud. From here you will be able to sample the power of generative ai in a scenario reflecting real-world usage.

Task 2. Using Prompt Engineering

In this lab we will interact with a Generative AI model to discuss . Don't worry, if you don't know much about this, use it as a learning exercise.

Generative AI is a type of artificial intelligence that can be used to create new information, such as text, images, or music. This makes it a powerful tool for learning about , as it can be used to generate realistic new content.

Generative AI can be used in a number of ways, including:

• Intelligent Chatbots for Conversation Practice: Generative AI-powered chatbots can mimic real conversations with native speakers. They can adapt to your skill level, offering targeted corrections and suggestions in real-time. This provides an immersive and low-stress environment to practice your speaking and listening skills.
• Customized Learning Paths: AI can analyze your strengths and weaknesses and suggest a personalized learning path. This might include grammar exercises, vocabulary lists, and reading passages tailored to your level and interests.
• Interactive Stories: Generative AI can create interactive narratives or role-playing scenarios where you engage with characters in a wide spectrum of situations. These stories can adapt based on your choices and language input, making you an active participant in a boundless world.
• Translating news or articles with AI assistance: Use AI to translate foreign language articles or other topics you are interested in. Compare the AI translation with the original for a deeper understanding of language structures and vocabulary in context.

Task 3. Context is Key

Generative AI Chat Bot needs context to answer questions. In the demo lab bot we have taken away some of its smarts to show the importance of context.

In the application we can put our questions to the AI model. To start, let's establish some basics around the importance of context relating to crafting questions.

1. Click on the persona

Note: Clicking on the chatbot persona will open the chat interface. From this interface you can interact with the generative ai model.

2. Start by saying hello to the chat bot, enter the following text

Hello Note: The chatbot will respond to this initial interaction with a helpful message to let you know it's ready to interact with you.

3. Let's start by asking a broad question to the AI

What is the current time in {{{ project_0.startup_script.lab_topic | "topic" }}}?

4. Let's try a more specific question

{{{ project_0.startup_script.lab_list | "Dynamic question" }}}?

5. Let's try to test the chatbots breadth of knowledge

What can you tell me about the {{{ project_0.startup_script.lab_item | "item" }}}?

Amazingly, we have just set the context for our journey. The Generative AI model has knowledge on lots of subjects. Here we are interested in finding out more about .

Task 4. Improving knowledge

Now that we have learned the basics, let's improve our knowledge of .

Note: Remember generative ai models are based on data taken from a point in time. The models need to be continually generated when working with time based information.

1. Let's ask a general question about the topic

Can you list three important facts about {{{ project_0.startup_script.lab_topic | "topic" }}} ?

2. Let's ask a specific question relating to an associated item

Are there any tourist destinations I should visit in {{{ project_0.startup_script.lab_topic | "topic" }}}

3. Lets ask a linking question

What's one thing you would recommend to do while in {{{ project_0.startup_script.lab_topic | "topic" }}}

4. Let's ask a broader question

Where is a good place to get food near the stadium in {{{ project_0.startup_script.lab_topic | "topic" }}}

5. Let's expand our query to generate some poetry

What teams are playing in {{{ project_0.startup_script.lab_topic | "topic" }}} Note: Want to learn more? Ask the chatbot more questions and see if it can help.

Great, it looks like you really know your stuff. Let's move onto the final assessment.

Task 5. Taking the Assessment

Finally, answer the assessment question based on .

Note: The lab assessment is accessed using the bottom navigation bar. Click on the Assess option to move to this screen. Answer correctly to pass the assessment. Once done, you can return from the chatbot application to the lab interface.

Click on the assess bottom navigation button to reveal the lab question. Select the correct answer to successfully complete the lab.

To successfully complete the lab assessment, each question presented must be answered correctly.

Note: The assessment validation button is available at the end of the question list. Once all questions are answered correctly, click this button to ensure the chatbot automatically allocates points in the running lab.

Click Check my progress to verify the objective.

Successfully complete the quiz assessment

Congratulations!

In just 30 minutes, you developed an understanding of Generative AI and the prompt engineering. You are now ready to take more labs.

What's next

• Try out other Google Cloud features for yourself. Have a look at our tutorials.
• Explore more Google Cloud solutions.
• Learn about the Firebase suite of products and services

Manual Last Updated May 16, 2024

Lab Last Tested May 16, 2024